Filling gaps of reference DNA barcodes in Syzygium from rainforest fragments in Sumatra

Given the difficulties for rapid biodiversity assessments in understudied regions, DNA barcoding appears as a suitable alternative. Still, this approach relies heavily on accurate reference sequence databases for correct taxonomic assignments. In this study, we evaluated the effectiveness of matK, rbcL, and ITS regions for the identification of Myrtaceae species with emphasis on the megadiverse genus Syzygium from Sumatra, Indonesia; and analyzed the applicability of species-tree inference for species assignment using barcode markers. ITS was the most variable barcode region (42.6% of variable sites), followed by matK (25.7%), and rbcL (14.9%). In terms of assignments of sequences using the BLAST algorithm, all markers were effective for genus-level attribution. For assignments at species rank, rbcL was able to attribute 30.15% of the samples at the species level, followed by matK (26.47%), and ITS (17.21%). These results are largely related to the availability of reference sequences for Myrtaceae in the databases since for the 27 species analyzed in this study, only 8 species had reference sequences for all three barcode regions available in GenBank. The species-tree inference based on the combination of matK, rbcL, and ITS markers recovered 41% of the species as monophyletic clades with strong node support. Due to its high level of differentiation, we recommend the ITS region as the most efficient barcode marker for the identification of Syzygium, and the traditional core-barcodes (matK + rbcL) as add-on barcodes.